considering multiple criteria for social categorization

Considering multiple criteria for social categorization can reduce intergroup bias Article

Accepted Version

final pre-publication version

Hall, N. R. and Crisp, R. J. (2005) Considering multiple criteriafor social categorization can reduce intergroup bias. Personality and Social Psychology Bulletin, 31 (10). pp. 1435-1444. ISSN 0146-1672 doi: https://doi.org/10.1177/0146167205276084 Available at https://centaur.reading.ac.uk/4486/

It is advisable to refer to the publisher’s version if you intend to cite from the work. See Guidance on citing .Published version at: http://dx.doi.org/10.1177/0146167205276084 To link to this article DOI: http://dx.doi.org/10.1177/0146167205276084

Publisher: SAGE Publications

All outputs in CentAUR are protected by Intellectual Property Rights law, including copyright law. Copyright and IPR is retained by the creators or other copyright holders. Terms and conditions for use of this material are defined in the End User Agreement .

www.reading.ac.uk/centaur

CentAUR

Central Archive at the University of Reading

http://centaur.reading.ac.uk/71187/10/CentAUR%20citing%20guide.pdf

http://www.reading.ac.uk/centaur

http://centaur.reading.ac.uk/licence

Reading’s research outputs online

1

This is the final pre-publication version of this article including all responses to referees comments.

The definitive version of this article can be found here DOI: 10.1177/0146167205276084

MS PSPB-03-075 Version 4

Running Head: MULTIPLE CATEGORIZATION

Considering multiple criteria for social categorization can reduce intergroup bias

Natalie R. Hall & Richard J. Crisp

University of Birmingham

Word count: 7237

Author Note

Natalie R. Hall, Richard J. Crisp, School of Psychology, University of Birmingham.

This research was funded by a British Academy research grant (SG-34045) to R. J. Crisp and

an Economic and Social Research Council doctoral studentship (PTA-030-2002-00889) to N. R.

Hall.

Correspondence concerning this article should be addressed to N.R. Hall or R. J. Crisp at the

School of Psychology, University of Birmingham, Edgbaston, Birmingham, B15 2TT. Tel: +44

(0)121 414 2227 or +44 (0)121 414 3335. Fax: +44 (0)121 414 4897. E-mail: [email protected] or

[email protected].

Multiple categorization 2

Abstract

Two experiments tested the notion that considering multiple criteria for social categorization

can reduce intergroup bias. In both experiments, participants were required to consider alternative

ways in which people could be classified, other than an initially salient intergroup dichotomy. In

Experiment 1 we found that generating alternative social classifications that were unrelated to an

initial target dichotomy reduced intergroup bias compared to a control condition. In Experiment 2

we replicated this effect and found that unrelated, but not related, categorizations were necessary to

reduce bias. This paper adds support to the view that increasing categorical complexity is a useful

tool in bias reduction. We discuss these findings in the context of a developing model of multiple

categorization effects.

Keywords: INTERGROUP BIAS, MULTIPLE SOCIAL CATEGORIZATION, IDENTITY

COMPLEXITY

Considering multiple criteria for social categorization can reduce intergroup bias


Processing resources are a valuable commodity for the social perceiver. For the most part,

classifying others into broad groupings -- social categorization -- is a useful shortcut for navigating

the social world and thus saving such resources (Fiske & Taylor, 1991; Hamilton, 1979; Macrae &

Bodenhausen, 2001). The usefulness of this mechanism can be demonstrated by the observation that

it is used even in artificial and arbitrary contexts (e.g., Rabbie & Horowitz, 1969; Tajfel, 1982;

Tajfel, Flament, Billig & Bundy, 1971; for a review see Bourhis, Sachdev, & Gagnon 1994). The

cognitive efficacy of social categorization comes, however, with some drawbacks: it is the mental

pre-requisite for intergroup bias. People in the perceiver’s own groups tend to be reliably and

consistently evaluated more positively than those in other groups (e.g., Brewer, 1979, Mullen,

Brown & Smith, 1992). In this article, we explore the potential for reducing such bias by

contravening the inherent ‘functionality’ of categorization.

The usefulness of social categorization

Categorization causes an accentuation of differences between, and similarities within, social

groups (Doise, 1978; McGarty & Penny, 1988; Tajfel & Wilkes, 1963). According to many theorists,

it is this process of differentiation between groups that, in part, leads to ingroup favouritism (for

examples see the Category Differentiation Model, Doise, 1978; Social Identity Theory, Tajfel &

Turner, 1979; Self-Categorization Theory, Turner, Hogg, Oakes, Reicher & Wetherall, 1987). There

are differences between these accounts. For some categorization is a simplifying mechanism, for

others a way of ascribing meaning to the social context, but they all rest on the notion that to

observe intergroup bias there must first exist a distinct cognitive dichotomization into ‘us’ and

‘them’ (see Maass & Schaller, 1989; Tajfel, 1969), and they are all linked by a common principle:

that categorization is on some level psychologically ‘useful’.


Social categorization can be considered useful because it is adaptive, providing a quick and

sufficient understanding of what is occurring in the social world (van Knippenberg & Dijksterhuis,

2000). Group-based judgments represent a slice of knowledge that can be applied to a target person

who fits the category (e.g., Sherman, Macrae, & Bodenhausen, 2000). Categorical thinking is

cognitively economical -- providing the maximum output of information with the least cognitive

effort (Macrae & Bodenhausen, 2001). It reduces the need to process incoming data by providing the

perceiver with group-based expectancies; and directs attention and processing resources to

expectancy relevant information (Hastie, 1981; Stangor & Macmillian, 1992). Using social categories

to guide processing therefore reduces the cognitive resources required for person perception by

providing a set of pre-conceived expectations about the target (Bodenhausen, 1992; Macrae, Stangor

& Milne, 1994; Oakes & Turner, 1990).

Although categorical thinking can be advantageous from this functional perspective,

stereotypes associated with category activation are often inaccurate when applied to individual

targets. Furthermore, such activated social categories can cause selective perception, interpretation,

and memory (Heilman, 1995). This can be problematic when the (sometimes negative) stereotypic

information generated is used to guide impression formation. The question we addressed in this

research was whether removing the cognitive efficacy of categorization would prevent its use in

guiding evaluative judgments in person perception. We suggest that one way to achieve this is by

making multiple alternative bases for social categorization salient for the perceiver.

Multiple social categorization

A considerable amount of work has focused on the notion that changing the categorical

representation of groups can correspondingly change intergroup evaluations. Levine and Campbell

(1972) drew attention to the reduced conflict associated with cultures having a crossed societal


structure. They noted that in contexts characterized by social structures that comprise loyalties to

multiple groups, some of which coincide, and some of which are in conflict, lower levels of

intergroup conflict tend to be observed. Later, this idea emerged in social psychological work on

intergroup relations. Deschamps and Doise (1978) suggested that by simultaneously combining two

cross-cutting dimensions of social categorization, opposing accentuation and assimilation principles

would effectively cancel each other out and reduce intergroup bias. This approach of combining two

cross-cutting categorizations has met with mixed success in terms of reducing intergroup bias

(Brown & Turner, 1979; Crisp, Hewstone, & Rubin, 2001; Deschamps & Doise, 1978;

Vanbeselaere, 1987; 1996). The majority of studies produce a pattern of evaluation where any group

associated with the ingroup is evaluated positively to some extent, but where negative evaluations

are still directed towards those people who are outgroup according to both categorical criteria

(Diehl, 1990; Ensari & Miller, 1998; Hewstone, Islam & Judd, 1993; Kenworthy, Canales, Weaver,

& Miller, 2003; Marcus-Newhall, Miller, Holtz & Brewer, 1993; Urada & Miller, 2000;

Vanbeselaere, 1987; see Crisp & Hewstone, 1999; Urban & Miller, 1999 for reviews).

In contrast to focusing participants on two cross-cutting dimensions of categorization

(Deschamps & Doise, 1978; Vanbeselaere, 1987; 1996), considering more than two categorization

criteria may uniquely lead to a ‘decategorization’ effect (Crisp et al., 2001). The reasoning is as

follows: we know that participants are cognitively able to use two dimensions of social

categorization under normal processing conditions (e.g., Vanbeselaere, 1987). As the number of

categorization dimensions to consider increases, the intergroup context would become much more

complex and the perceiver may no longer be able to easily determine on what basis they are similar

to, or different from, the target person. There may be too many dimensions of membership to allow

meaningful category conjunctions to form. Perceivers may be able to cognitively combine categories


like ‘single mother’ or ‘working parent’ (see Hutter & Crisp, in press), but they may find, for

example, ‘single, black, educated, working mother’ more difficult, at least with respect to gaining

any added value by processing information and making evaluative judgments heuristically. There

may come a point in considering successive categorical dimensions when using a heuristic method of

classification actually becomes inefficient relative to other methods of impression formation. The

idea of multiple modes of processing is key to contemporary theories of person perception, and we

suggest that the ideas outlined above regarding the effects of multiple categorization can be

interpreted within the context of some well-established frameworks relating to impression

formation.

Categorization and decategorization

Brewer’s (1988) Dual Process Theory and Fiske and Neuberg’s (1990) Continuum Model

both consider person perception and impression formation to comprise of two distinct processes:

categorization and individuation. The basic premise of Brewer’s Dual Process Theory (1988) is that

either a top-down category based approach or a bottom-up person based approach is used and this is

analogous to a heuristic (category) versus systematic (individuated) approach to person perception.

Fiske and Neuberg (1990) propose a continuum for impression formation where one extremity is

category-based processing and the other is attribute (individuated) based processing. On this

continuum people can be perceived as a representative of a group (category-based responding), or as

an individual separate from any categorical identity (attribute-based responding). Fiske and Neuberg

propose a system where an initial categorization will occur, followed by a confirmatory process to

determine how good a fit there is between the category and target. If this is unsuccessful, then there

will be a shift towards the individuated mode of perception by invoking an attribute-by-attribute

approach to form an impression of the target person. As Fiske, Lin and Neuberg (1999) state:


“...perceivers typically use category-oriented processes before they use attribute-orientated processes,

and [that] if the category-oriented processes work well enough perceivers do not engage additional

more attribute oriented processes” (pp. 236). Brewers’ Dual Process Theory is based on similar

principles and predicts that the mode of perception will change from the use of category based

heuristics to a more systematic individuated mode of perception under conditions that either favor,

or do not favor, one over the other. As Brewer and Harasty Feinstein (1999) note, the changes in

processing outlined by the Dual Process Model map directly onto Fiske and Neuberg’s impression

formation continuum.

This switch in processing from using categorization to individuation can be termed

decategorization. Decategorization results when initial categorizations of people as group members

are overridden by a more comprehensive and differentiated perception of the individual (see Brewer

& Miller, 1984; Ensari & Milller, 2001; Wilder, 1986). Once decategorization has occurred the target

person should be primarily defined as an individual rather than a group member, which should

remove category-based bias. Previous research has found that conditions that promote

decategorization will encourage participants to favour the ingroup less (presumably because the

judgment is being made on an appreciation of personal merit, not on the basis of pre-conceived

categorical expectancies; see Brewer & Miller, 1984; Krueger & Rothbart, 1988; Miller &

Harrington, 1992). This change in the mode of processing should also allow the perceiver to develop

a more personalized and less homogenous perception of ingroup and outgroup members (Ensari &

Miller, 2001; Fiske & Neuberg, 1989).

We argue that considering multiple criteria for social categorization could produce a

decategorization effect by virtue of the implied decrease in usefulness of any one dimension of social

comparison. Decategorization is, however, an uncommon finding when only two dimensions of


categorization are combined. Ensari and Miller (2001) directly compared personalized contact

(where participants are asked to consider unique attributes) with the traditional crossed

categorization approach (two dimensions of categorization). They found that intergroup bias was no

longer moderated by shared ingroup memberships in the personalised contact condition, but this

was not the case in the crossed categorization condition. This could be because participants are quite

able (and appear to) consider two social categorizations simultaneously in memory and use them to

guide evaluative judgments (Vanbeselaere, 1987; 1996).

However, as the number of category dimensions that need to be processed simultaneously

increases then it may be that categorization, as a functional heuristic, becomes relatively inefficient,

leading to a shift in orientation and a reduction in bias as outlined by Brewer’s (1988) and Fiske and

Neuberg’s (1990) models. If perceivers experience conditions in which categorization becomes too

complex to be functional then they may resort to a more individuated strategy for person

perception. Correspondingly they should be less biased in their evaluations of ingroup and outgroup

members, as these evaluations are no longer tied to group membership. By considering multiple

criteria for social categorization, category-oriented processes may no longer provide the most

functionally efficient or meaningful way of making judgments, and consequently perceivers may

resort to a less biased approach to person perception.

In the context of the dual processing accounts of impression formation outlined above, we

can make some clear predictions with respect to the effects of multiple categorization on intergroup

evaluations. We thus tested the possibility that merely thinking about multiple alternative criteria

for social categorization may decategorize person perception and reduce category-based ingroup

favoritism.

Pilot study


We initially carried out a pilot study to develop our multiple categorization task. We focused

on undergraduate students’ own university group membership versus a local rival university. Forty-

five undergraduate psychology students took part in individual sessions. All participants were

informed that they would be asked to think about people who shared their university membership

and people who did not. They were advised that the study was concerned with perceptions of

people at different universities and how their respective qualities influence creativity and originality

in working environments, and that all information would be compiled to assess the skills and

abilities from graduates at different universities. Participants first completed a task that was

ostensibly designed to test creativity, originality, and artistic merit. They were provided with a

square piece of paper and asked to make a model that represented something in the real world by

folding or tearing. They were allowed five minutes to complete this task. Participants in the

experimental condition were then given the following instructions that constituted the critical task

variable manipulation: “On an earlier page we asked you to think about your university affiliation.

This is one possible way that people can define themselves, but there are many more. People can be

similar or different from others not just according to one criterion, but often there are many and

varied ways that people define themselves and others. We would like you to think of other groups

that university students can use to define themselves (in addition to what university they are at)”.

They were asked to provide a few written lines of elaboration on each alternative they came

up with to reinforce the manipulation. They then completed a measure of group evaluation.

Participants in the control condition did not complete a categorization task. Instead, they

completed the group evaluations directly after making their paper model and received none of the

instructions outlined above.


Evaluations were measured by participants rating two paper models supposedly made by

previous ingroup and outgroup participants1

. Participants were asked to rate an ingrouper’s and

outgrouper’s paper model according to creativity, originality and artistic merit on scales ranging

from 0 (not at all) to 9 (very much so). These items were averaged to form indices of group

evaluation: ingroup =.721, outgroup =.722. The group to which the model was affiliated and the

order in which the models were presented was counterbalanced across conditions.

A 2 (task: control vs. multiple alternatives) x 2 (target: ingroup vs. outgroup) mixed

ANOVA with repeated measures on second factor revealed a significant interaction between task

and target F (1, 43) = 4.77, p = .035. This interaction reflected significant variation in intergroup

bias (ingroup minus outgroup evaluations) as a function of task instructions, although not in the

expected direction. Compared to the control condition (M= -.300), intergroup bias was increased

following the generation of multiple alternatives (M = .707)2

.

To investigate why we did not obtain the predicted reduction in bias we examined the

specific alternative classifications generated by our participants in the experimental condition. We

found that the majority of participants who completed the multiple categorization task (78%) had

actually produced classifications that were related in some way to the initial target dichotomy, rather

than unrelated alternative ways of classifying the student groups. We define related category

dimensions as those that are conceptually linked with the target categories. For example, what

academic subject [major] is being studied is related to the context within which the target intergroup

dichotomy (in this case ‘students’) is embedded. In contrast, unrelated category dimensions are those

that are not conceptually linked with the target categories. For example, in our studies ‘nationality’

is unrelated to being a student. In this pilot study most participants had generated category

dimensions like ‘subject [major] studied’ or ‘hall or residence [fraternity/sorority]’ as their


alternative criteria. These alternatives are, however, directly related to the dimension ‘student

affiliation’. Few participants generated alternatives that were (as we had intended) unrelated to the

target intergroup context (e.g., ‘gender’ or ‘place of birth’). Interestingly, closer examination of the

participants who had generated unrelated criteria did show a trend in the predicted direction

(reduced bias).

We hypothesized that generation of related alternatives may not undermine the usefulness of

the target dichotomy. Our reasoning was as follows. We know from previous work that whilst

making a superordinate categorization salient can sometimes reduce intergroup bias (Gaertner &

Dovido, 2000), sometimes it can be ineffective, or even increase bias (see Hornsey & Hogg, 2000). In

their Ingroup Projection Model Mummendy and Wenzel (1999) draw upon Self-Categorization

Theory (Turner, 1981; Turner, Hogg, Oakes, Reicher & Weatherall, 1987) to argue that making a

superordinate categorization salient can accentuate intergroup bias by introducing a relevant basis

for social comparison. For example, fans of opposing sports teams may not be concerned with

making ingroup favoring social comparisons against some other team until a relevant superordinate

dimension is salient (i.e., they become rivals within the same league or competition).

We argue that the generation of related alternative categorizations may have similar effects.

Thinking of related categorizations, by definition, should prime a relevant superordinate category

dimension (‘students’) along which to compare the initial ingroup and outgroup student subgroups

(defined by university affiliation). For example, categories related to university affiliation such as

fraternity or sorority, or major studied, all make the superordinate category ‘students’ more salient

and relevant. Put another way, thinking about bases for categorization other than the target

dichotomy, but nonetheless related to it, creates a context of comparability in which ingroup

favoring evaluations can serve to positively differentiate the ingroup from the outgroup (Tajfel &


Turner, 1979; see also Hogg & Abrams, 1990; Jetten, Spears & Manstead, 1996; 2001). Generation of

related alternatives reinforces any potential comparability between the ingroup and outgroup, thus,

ingroup favoritism should at best, remain pervasive, and at worse increase.

This pilot study therefore established an important pre-condition for the generation of

multiple alternative classifications to reduce intergroup bias. To avoid priming a superordinate

context of comparison as outlined above, it would be important to specify the generation of

unrelated alternatives. In our first experiment, we therefore modified our instructions to achieve this

goal.

Experiment 1

In this experiment the multiple categorization task instructions used in the pilot study were

adapted to encourage participants to generate unrelated, not related, alternatives. We predicted that

relative to a control condition, generation of unrelated alternatives would result in a decrease in

intergroup bias.

Method

Participants and Design

Thirty-eight undergraduate psychology students (31 females, 7 males, with an overall mean

age of 19) were randomly allocated to a 2 (task: control vs. multiple unrelated alternatives) x 2

(target: ingroup vs. outgroup) mixed design with repeated measures on the second factor.

Participants received course credit for their involvement.

Procedure

The procedure was identical to the pilot study except that after having created the paper

models, participants were given the multiple categorization instructions outlined above with the

addition of an explicit instruction not to generate related alternatives (the student subgroups ‘subject


studied ’ and ‘hall of residence’ were given as examples of alternatives to avoid). Participants were

asked instead to generate “alternative ways people can vary independent from being a student”, with

the example given of ‘Nationality’.

Participants in the control condition did not carry out the multiple categorization task and

simply completed the evaluative measures (the typical baseline condition used in bias-reduction

research) using the same paper models as employed in the pilot study.

Dependant Measures

Evaluations were measured by participants rating two paper models for creativity on a scale

ranging from 0 (not at all) to 9 (very much so)3

. One of the models ostensibly belonged to an

ingroup member and one ostensibly belonging to an outgroup member. The group to which the

model was affiliated and the order in which the models were presented was counterbalanced across

conditions.

Results & discussion

One participant did not fully complete the dependant measures so was excluded from all

further analysis. Four participants were removed on the grounds that they were suspicious about the

aims of the experiment. Despite the refined multiple categorization instructions one participant

exclusively produced related classifications so they were also excluded leaving an N of 32, with the

remaining participants evenly distributed across conditions. Means and standard deviations can be

found in Table 1.

Manipulation check

The categories generated by participants in the unrelated alternatives condition were judged

for relatedness by two independent raters blind to the experimental hypotheses (participants in the

control condition did not produce any categories). The judges were asked to rate the five category


dimensions produced by each participant on a seven point scale as to whether they were unrelated

(anchored at 1) or related (anchored at 7) to the initial target category dimension (student affiliation).

Judges were provided with an example to clarify this distinction. An example of an unrelated

category was ‘hair color’, an example of a related category was ‘subject studied’. The two blind

coders’ ratings had acceptable inter-rater reliability, r = .688, p =.002, (MJudge 1 = 1.41, SDJudge 1 = .618

vs. MJudge 2 = 2.12, SDJudge 2 = 1.50) so were collapsed into one scale (by taking the mean of the two

judges’ ratings). A one sample t-test against the scale midpoint (4) revealed that on aggregate

participants in the experimental condition had produced category dimensions that were unrelated to

the target dimension of comparison (‘students’), t (16) = -9.35, p < .0005 (M = 1.76, SD = .986).

Evaluations

A 2 (task) x 2 (target) mixed ANOVA with repeated measures on the second factor revealed no

main effect of target, F (1, 30) = .868, p = .359, nor of task, F (1, 30) = 2.41, p = .131. There was,

however, the predicted interaction between task and target, F (1, 30) = 7.57, p =.010. In line with

expectations, intergroup bias (ingroup minus outgroup evaluations) was lower following the

generation of multiple unrelated alternative categorizations (M = -.083) compared to the control

condition (M = 1.67)4

.

In summary, in line with predictions, intergroup bias was reduced following the generation of

multiple unrelated categorizations compared to the control condition. The effect appeared to be

driven predominantly by lowered ingroup evaluations, although some (non-significant) variation in

outgroup evaluations was also observed. We know that intergroup bias is driven primarily by

accentuated ingroup evaluations rather than outgroup derogation (Brewer, 1979). As such, reduced

reliance on social categorization should, as we observed, result in a decrease in this ingroup

favoritism (which is driving the bias in the first place), rather than a decrease in outgroup


derogation. In Experiment 2 we sought to replicate and expand on these findings by directly

comparing the effects of generating related versus unrelated alternative bases for categorization.

Experiment 2

Our pilot work indicated that thinking about related alternative classifications had a

detrimental effect on intergroup relations, yet, we did not experimentally test this in Experiment 1.

In Experiment 2 we therefore directly compared the two ways of framing the task. We had decided

to use only a single-item evaluative measure in Experiment 1 so we expanded this to a three-item

measure in this experiment. We expected reduced intergroup bias only for participants who were

instructed to generate unrelated alternatives, compared to a control condition and a condition in

which participants generated related alternatives.

Method

Participants and Design

Fifty-eight undergraduate psychology students (56 females; 2 males with a mean age of 19)

were randomly allocated to a 3 (task: control vs. multiple unrelated alternatives vs. multiple related

alternatives) x 2 (target: ingroup vs. outgroup) mixed design with repeated measures on the second

factor. Participants took part in exchange for course credit.

Procedure

The procedure was the same as in Experiment 1 except we added a related alternatives

condition. This version of the multiple categorization task comprised the same core instructions

used in the pilot study but instead of the addition of an explicit instruction not to generate related

alternatives, participants were specifically instructed to generate related alternatives (the student


subgroups ‘subject area’ and ‘hall of residence’ were given as examples of alternatives to use). The

‘unrelated’ task was identical to that used in Experiment 1.

Dependent measures

Participants were asked to rate the ingroup and outgroup paper models according to

creativity, originality and artistic merit on scales ranging from 0 (not at all) to 9 (very much so).

These items were averaged to form indices of ingroup ( = .762) and outgroup ( = .810) evaluation

respectively.

Results & discussion

Three participants were removed because they reported suspicion about the aims of the

experiment, leaving an N of 55. Means and standard deviations for all measures can be found in

Table 2.

Manipulation check

The categories generated by participants in the related and unrelated alternatives conditions

were judged by the same two independent raters, blind to condition, as described in Experiment 1.

We also used the same ‘relatedness’ scale for judging participants’ responses as used in Experiment 1.

The ratings produced by the two blind coders had acceptable inter-rater reliability, r = .905, p <

.001 (unrelated: MJudge 1 = 1.20, SDJudge 1 = .696 vs. MJudge 2 = 1.85, SDJudge 2 = 1.18, related: MJudge 1 =

4.44, SDJudge 1 = 1.58 vs. MJudge 2 = 5.56, SDJudge 2 = 1.89) so were collapsed (by taking the mean) into

one index. On this relatedness index, a t-test revealed a significant difference between related and

unrelated conditions in the expected direction, t (36) = -8.30, p < .0005, (Munrelated = 1.52 vs. Mrelated =

5.00). Participants in the related condition generated categories that were more related to the student

superordinate category context than those in the unrelated condition. One sample t-tests were

carried out against the midpoint of the scale. Those in the related condition produced categories that


were judged to be significantly related to the superordinate context, t (17) = 2.58 p = .020. In

contrast, those in the unrelated condition produced categories that were judged to be significantly

unrelated to the superordinate context, t (19) = -13.01, p < .0005.

Evaluations

A 3 (task) x 2 (target) mixed ANOVA with repeated measures on the second factor revealed

no main effect of target, F (1, 52) = .004, p = .950, nor of task, F (1, 52) = .261, p = .771. There was,

however, the expected significant interaction between task and target, F (2, 52) = 3.84, p = .028.5

To

explore the observed interaction, we again created an index of intergroup bias (ingroup minus

outgroup evaluations). Consistent with the findings from Experiment 1, planned comparisons

revealed less intergroup bias in the unrelated alternatives condition than in the control condition, t

(52) = 2.23 p = .030 (Mcontrol = .249, vs. Munrelated = -.650). There was also less intergroup bias in the

unrelated alternatives condition compared to the related alternatives condition, t (52) = 2.49, p =

.016 (Munrelated = -.650, vs. Mrelated = .338). Finally, there was no difference in intergroup bias between

the control and the related alternatives condition, t (52) = -.219, p = .828, (Mcontrol = .249, vs. Mrelated

= .338).

To summarize, in Experiment 2 we supported the idea that merely thinking about multiple

alternative categorizations can reduce intergroup bias -- as long as they are unrelated to the initial

intergroup context.

General Discussion

The aim of these experiments was to see if considering multiple alternative categorizations to

the target ingroup versus outgroup categorization could reduce intergroup bias. In Experiment 1 we

found that the generation of multiple unrelated alternatives led to the predicted bias-reduction

effect. Experiment 2 built on these findings by replicating the decrease in intergroup bias following


generation of multiple unrelated, but not related, categorization criteria. Overall, these studies have

developed and refined a new approach to bias-reduction that is consistent with extant theoretical

and empirical work on dual processing accounts of impression formation.

Multiple alternatives and decategorization

In Experiments 1 and 2 we found that the generation of multiple criteria for social

categorization that were unrelated to the superordinate category comparison context ‘students’

reduced intergroup bias. The findings are consistent with the suggestion that the generation of

multiple alternatives can encourage a shift in judgment orientation: if categorization is no longer

functional in such contexts then judgments become less category-based (Brewer, 1988; Fiske &

Neuberg, 1990), reducing ingroup favoritism.

There is an important caveat to this approach. Although the generation of multiple

categories can reduce ingroup favoritism, it will only do so if the categories generated are unrelated

to the superordinate category context. We believe that this is a crucial prerequisite for ensuring the

success of multiple categorization approaches that aim to achieve decategorization. The generation

of related categories may emphasize ingroup and outgroup comparability by highlighting the

superordinate category context. Correspondingly, this may have no bias-reducing effect, or even

result in a reactive increase in intergroup bias.

There are a number of potential explanations for what might cause the observed reduction in

bias. We have argued that the generation of multiple categories may undermine the utility of the

initial dichotomy and hence prevent the application of stereotypes relating to initial ingroups and

outgroups. It is also possible, however, that producing numerous categorizations may reduce bias

along the specified ingroup-outgroup categorization by reducing the degree to which the categorical


distinction is salient for the perceiver. The decategorization effect may also be caused by a dilution

of the original groups and their associated characteristics with many other categorizations and their

stereotypes. The aim of the current research was to establish whether generation of multiple

alternative categorizations can lead to a reduction in bias. Having now established this, more

research is needed to delineate precisely the mechanisms that underlie this effect.

We should acknowledge that some aspects of our findings should be considered tentative. We

have demonstrated the effect on only one dependent measure and with one paradigm thus, it is

necessary for future work to test generalizability of the findings. In addition, while the general

pattern of the findings across studies is consistent with predictions, there are also a number of

specific inconsistencies across the three studies (relating to whether the locus of effect resides in

moderation of ingroup evaluations, outgroup evaluations, or both). This further strengthens the call

for replication with other procedures and measures.

Category complexity and ingroup projection

The multiple categorization approach used in this research may relate to the Ingroup

Projection Model, Mummendy and Wenzel (1999). This model builds upon Self-Categorization

Theory (Turner, 1981; Turner et al., 1987) to argue that superordinate categorization creates a

context of comparability between ingroup and outgroup subgroups. Mummendey and Wenzel argue

that in intergroup situations where the superordinate group is salient, the ingroup prototype will be

projected on to the superordinate group. This ingroup projection may cause the outgroup to be

perceived more negatively. If both the ingroup and outgroup are conceptualized as part of a

superordinate category (a relevant comparison context), and the ingroup prototype has been

projected onto the superordinate category, then the outgroup will be seen as deviating from the

norm (Wenzel, Mummendey, Weber, & Waldzus, 2003). The ingroup projected prototype will


invariably be positive and hence the outgroup will be judged less positively by comparison (Wenzel,

2001). We argue that the generation of category dimensions related to the target dichotomy primes

the superordinate categorization context, making the ingroup and outgroup similarly comparable.

In contrast, generating unrelated alternatives reduces bias because it creates a context in which the

utility of categorization per se is reduced – the context becomes too complex to usefully employ

social categories.

Interestingly, the notion that category complexity can reduce intergroup bias is also reflected

in the Ingroup Projection Model. Waldzus, Mummendey, Wenzel, and Weber (2003) found that

projection of the ingroup prototype does not occur when the superordinate category is perceived

itself as having a complex prototype. Although we argue that generating unrelated alternatives

should lead to categorization being abandoned as a means of structuring social perception (so should

not invoke the superordinate category), there are clear parallels between our model and the Ingroup

Projection Model, in that both argue that complexity can reduce intergroup bias. Indeed, the notion

that complexity is a good thing for intergroup relations is emerging as a key theme in work on

intergroup relations. Roccas and Brewer’s (2002) model of Social Identity Complexity argues that

the more complex a perceiver’s social identity, and the less overlap there is between their multiple

ingroup identities, the less ingroup favoritism should be observed. Although not specific to

ingroups, our research supports this general concept.

The findings we report reiterate the message that multiple categorization can reduce

intergroup bias. Future research could examine some interesting possibilities regarding the

integration of multiple categorization and ingroup projection via notions of complexity, and the

corresponding potential for developing effective interventions to reduce intergroup bias.

Conclusions


The aim of this research was to explore the potential for multiple social categorization to

reduce intergroup bias. Could merely thinking about multiple alternative bases for social

categorization be an effective strategy for promoting more egalitarianism in person perception? On

the basis of these findings, we argue that the answer is yes -- but with one important caveat. The way

in which people are asked to consider multiple alternative criteria is critical to the observed effects.

We found that categorization no longer defined evaluative judgment following multiple

categorization providing that the categories considered were not related to the superordinate

category context. These findings provide some important links between diverse literatures in

impression formation and person perception, but perhaps more importantly, may also go some way

to developing an integrated model of the implications of multiple categorization for intergroup

relations.

References

Bodenhausen, G. V. (1992). Information processing functions of generic knowledge

structures and their role in context effects in social judgement. In N. Schwartz & S. Sudman (Eds.),

Context effects in social and psychological research (pp.267-278). New York: Springer Verlag.

Bourhis, R. Y., Sachdev, I., & Gagnon, A. (1994). Intergroup research with the Tajfel

matrices: Methodological notes. In M. P. Zanna & J. P. Olson (Eds.), The psychology of prejudice: The

Ontario symposium, (pp. 209-232). New York: Random House.

Brewer, M. B. (1979). Ingroup bias in the minimal intergroup situation: A cognitive-

motivational analysis. Psychological Bulletin, 86, 307-324.


Brewer, M.B. (1988). A dual process model of impression formation. In T. K. Srull & R. S.

Wyer Jr (Eds.) Advances in Social cognition (pp. 1-36). Erlbaum: Hillsdale, NJ.

Brewer, M. B., & Harasty Feinstein, A. S. (1999). Dual processes in the cognitive

representation of persons and social categories. In S. Chaiken & Y. Trope (Eds.) Dual process theories

in social psychology (pp. 255-270). New York: Guilford.

Brewer, M. B., & Miller, N. (1984). Beyond the contact hypothesis: Theoretical perspectives

on desegregation. In N. Miller & M. B. Brewer (Eds.), Groups in contact: The psychology of

desegregation (pp. 281-302). San Diego, CA: Academic Press.

Brown, R. J., & Turner, J. C. (1979). The criss-cross categorization effect in intergroup

discrimination. British Journal of Social & Clinical Psychology, 18, 371-383.

Crisp, R.J., & Hewstone, M. (1999). Differential evaluation of crossed category groups:

Patterns, processes, and reducing intergroup bias. Group Processes & Intergroup Relations, 2, 307-333.

Crisp, R. J., Hewstone, M., & Rubin, M. (2001). Does multiple categorization reduce

intergroup bias? Personality & Social Psychology Bulletin, 27, 76-89.

Deschamps, J. C., & Doise, W. (1978). Crossed category memberships in intergroup

relations. In H. Tajfel (Ed.), Differentiation between social groups (pp. 141-158). Cambridge:

Cambridge University Press.

Diehl, M. (1990). The minimal group paradigm: Theoretical explanations and empirical

findings. In W. Stroebe & M. Hewstone (Eds.), European Review of Social Psychology, (Vol.1, pp. 263-

292). Chichester: Wiley.

Doise, W. (1978). Groups and individuals. Explanations in Social Psychology, Cambridge:

Cambridge University Press.


Ensari, N., & Miller, N. (1998). Effect of affective reactions by the outgroup on preferences

for crossed categorization discussion partners. Journal of Personality & Social Psychology, 75, 1503-

1527.

Ensari, N., & Miller, N. (2001). Decategorization and the reduction of bias in the crossed

categorization paradigm. European Journal of Social psychology, 31, 193-216.

Fiske, S. T., Lin, M., & Neuberg, S. L. (1999). The continuum model: ten years later. In S.

Chaiken & Y. Trope (Eds.) Dual process theories in social psychology (pp. 231-254). New York:

Guilford.

Fiske, S. T., & Neuberg, S. L. (1989). Category-based and individuating processes as a

function of information and motivation: Evidence from our laboratory. In D. Bar-Tal, C. F.

Graumann, A. W. Kruglanski, & W. Stroebe (Eds.), Stereotyping and Prejudice: changing conceptions

(pp. 83-104). Springer-Verlag: New York.

Fiske, S. T., & Neuberg, S. L. (1990). A continuum of impression formation, from category-

based to individuating processes; Influences of information and motivation on attention and

interpretation. In M. P. Zanna (Ed.), Advances in Experimental Social Psychology (Vol. 23, pp. 1-74).

New York: Academic Press.

Fiske, S. T., & Taylor, S. E. (1991). Social Cognition. New York: McGraw Hill.

Gaertner, S. L., & Dovidio, J. F. (2000). Reducing intergroup bias: The Common Ingroup Identity

Model. Philadelphia, PA: The Psychology Press/Taylor & Francis.

Hamilton, D. L. (1979). A cognitive attributional analysis of stereotyping. In L. Berkowitz

(Ed.), Advances in Experimental Social Psychology (Vol. 12 pp.53-84). New York: Academic Press.


Hastie, R. (1981). Schematic principles in human memory. In E. T. Higgins, C. P. Herman,

& M. P. Zanna (Eds.), Social cognition: The Ontario Symposium (Vol. 1, pp. 39-88). Hillsdale, NJ:

Erlbaum.

Heilman, C. M. (1995). Sex stereotypes and their effects in the workplace: What we know

and what we don’t know. Journal of Social Behaviour and Personality, 10, 3-26.

Hewstone, M., Islam, M. R., & Judd, C. M. (1993). Models of crossed categorizations and

intergroup relations. Journal of Personality & Social psychology, 64, 779-793.

Hogg, M. A., & Abrams, D. (1990). Social motivation, self-esteem and social identity. In D.

Abrams & M. A. Hogg (Eds.), Social identity theory: Constructive and critical advances (pp. 28-47).

New York: Harvester Wheatsheaf.

Hornsey, M. J., & Hogg, M. A. (2000). Intergroup similarity and subgroup relations: Some

implications for assimilation. Personality & Social Psychology Bulletin, 26, 948-958.

Hutter, R. R. H., & Crisp, R. J. (in press). The composition of category conjunctions.

Personality and Social Psychology Bulletin.

Jetten, J., Spears, R., & Manstead, A. S. R. (1996). Intergroup norms and intergroup

discrimination: Distinctive self-categorization and social identity effects. Journal of Personality &

Social Psychology, 71, 1222-1233.

Jetten, J., Spears, R., & Manstead, A. S. R. (2001). Similarity as a source of differentiation:

the role of group identification. European Journal of Social Psychology, 31, 621-640.

Kenworthy, J. B., Canales, C. J., Weaver, K. D., & Miller, N. (2003). Negative incidental

affect and mood congruency in crossed categorization. Journal of Experimental Social Psychology, 39,

195-219.


Krueger, J., & Rothbart, M. (1988). Use of categorical and individuating information in

making inferences about personality. Journal of Personality & Social Psychology, 55, 187-95.

Levine, R. A., & Campbell, D.T. (1972). Ethnocentrism: Theories of conflict, ethnic attitudes

and group behaviour. New York: Wiley.

Maass, A., & Schaller, M. (1989). Illusory correlation and social categorization- toward an

integration of motivational and cognitive factors in stereotype. Journal of Personality & Social

Psychology, 56, 709-721.

Macrae, C. N., & Bodenhausen, G. V. (2001). Social Cognition: Categorical person

perception. British Journal of Psychology, 92, 239-255.

Macrae, C. N., Stangor, C. & Milne, A. B. (1994). Activating social stereotypes: A functional

analysis. Journal of Experimental Social Psychology, 30, 370-389.

Marcus-Newhall, A., Miller, N., Holtz, R., & Brewer, M. B. (1993). Cross-cutting category

memberships with role-assignment: a means of reducing intergroup bias. British Journal of Social

Psychology, 32, 125-146.

McGarty, C., & Penny, R. E. C. (1988). Categorization, accentuation and social judgement.

British Journal of Social Psychology, 27, 147-157.

Miller, N., & Harrington, H. J. (1992). Social categorization and intergroup acceptance:

Principles for the design and development of cooperative learning teams. In R. Hertz-Lazarowitz &

N. Miller, (Eds.), Interaction in cooperative groups: the theoretical anatomy of group learning (pp. 203-

227). Cambridge University Press; New York.

Mullen, B., Brown, R. J., & Smith, C. (1992). Ingroup bias as a function of salience, relevance

and status: an integration. European Journal of Social Psychology, 22 103-122.


Mummendey, A. & Wenzel, M. (1999). Social discrimination and tolerance in intergroup

relations: Reactions to intergroup difference. Personality and Social Psychology Review, 3, 158-174.

Oakes, P. J., & Turner, J. C. (1990). Is limited information processing capacity the cause of

social stereotyping? In W. Stroebe & M. Hewstone (Eds.), European Review of Social Psychology

(Vol. 1, pp. 111-135). Chichester: Wiley.

Rabbie, J. M., & Horowitz, M. (1969). Arousal of ingroup-outgroup bias by chance win or

loss. Journal of Personality & Social Psychology, 13, 269-277.

Roccas, S. & Brewer, M. B. (2002). Social identity complexity. Personality and Social

Psychology Review, 6, 88-106.

Sherman, J. W., Macrae, C. N., & Bodenhausen, G. V. (2000). Attention and stereotyping:

Cognitive constraints on the construction of meaningful social impressions. European Review of

Social Psychology (vol. 11, pp. 145-175). Chichester: John Wiley & Sons Ltd.

Stangor, C., & Macmillian, D. (1992). Memory for expectancy-congruent and expectancy-

incongruent information: A review of the social and social developmental literatures. Psychological

Bulletin, 111, 42-61.

Tajfel, H. (1969). Cognitive aspects of prejudice. Journal of Social Issues, 25, 79-97.

Tajfel, H. (1982). Social psychology of intergroup relations. Annual Review of Psychology, 33,

1-39.

Tajfel, H., Flament, C., Billig, M., & Bundy, R. F. (1971). Social categorization and

intergroup behaviour. European Journal of Social Psychology, 1, 149-178.

Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In W. G.

Austin & S. Worchel (Eds.), The Social Psychology of Intergroup Relations (pp. 33-47). California:

Brooks/Cole.


Tajfel, H., & Wilkes, A. (1963). Classification and quantitative judgement. British Journal of

Social Psychology, 54, 101-114.

Turner, J. C. (1981). The experimental social psychology of intergroup behaviour. In J. C.

Turner & H. Giles (Eds.). Intergroup Behaviour (pp. 66-101). Oxford: Basil Blackwell.

Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Re-

discovering the Social Group: A Self-categorization Theory. Oxford: Blackwell.

Urada, D. I., & Miller, N. (2000). The impact of positive mood and category importance on

crossed categorization effects. Journal of Personality and Social Psychology, 78, 417-433.

Urban, L. M., & Miller, N. (1998). A theoretical analysis of crossed categorization effects: A

meta-analysis. Journal of Personality and Social Psychology, 74, 894 908.

van Knippenberg, A. V., & Dijksterhuis, A. (2000). Social categorization and Stereotyping: a

functional perspective. European Review of Social Psychology (vol. 11, pp. 105-144). Chichester: John

Wiley & Sons Ltd.

Vanbeselaere, N. (1987). The effect of dichotomous and crossed categorizations on

intergroup discrimination. European Journal of Social Psychology, 17, 143-456.

Vanbeselaere, N. (1996). The impact of differentially valued and overlapping categorizations

upon the differentiation between positively, negatively and neutrally evaluated social groups.

European Journal of Social Psychology, 26, 75-96.

Waldzus, S. Mummendey, A. Wenzel, M. & Weber, U. (2003). Towards tolerance:

Representations of superordinate categories and perceived ingroup prototypicality. Journal of

Experimental Social Psychology, 39, 31-47.


Wenzel, M. (2001). A social categorization approach to distributive justice: Social identity as

the link between relevance of inputs and need for justice. British Journal of Social Psychology, 40, 315-

335.

Wenzel, M., Mummendey, A., Weber, U., & Waldzus, S. (2003). The ingroup as pars pro

toto: projection from the ingroup onto the inclusive category as a precursor to social discrimination.

Personality and Social Psychology Bulletin, 29, 461-473.

Wilder, D. A. (1986). Social Categorization: Implications for creation and reduction of

intergroup bias. In L. Berkowitz (Ed.), Advances in Experimental Social Psychology, (Vol. 19 pp.293-

355).


Endnotes

1

The dependent measures and models used were extensively pre-tested. The two models presented

were of a boat and a crown. These were originally made in a pre-testing session with a group of

students. Participants were informed that for the purposes of the pre-test the models would be

referred to as ‘model 1’ and ‘model 2’. Participants were required to rate model 1 and model 2 in

terms of originality, creativity and artistic merit. For the boat; originality t (13) = .069, p = 0.95,

creativity t (13) = -.623, p = .544, and artistic merit t (13) = .203, p = .842. For the crown;

originality t (13) =.648, p =.528, creativity t (13) = 1.27, p =.226 and artistic merit t (13) = 1.72, p =

.110. Which model was presented first as ‘model 1’ was counterbalanced across participants. The

order of presentation of the models in the pre-test had no effect on participants’ ratings of

originality, creativity or artistic merit for either model. We combined the three scales into one

overall measure for each model (boat Cronbach’s = .705; crown Cronbach’s =.655). When no

reference was made to ingroup or outgroup affiliation of the person who made the model, both

models were perceived to be equally positive, t (14) = -1.20, p = .251. We thus used these models

and the three evaluative items in the subsequent experiment.

2 One way t-tests were also computed against the scale midpoint of 0 (which would indicate no

bias). There was no evidence of bias in the control condition, t (19) = -.933, p = .363, but there was

significant ingroup favoritism in the multiple alternatives condition, t (24) = -2.19, p = .038.

3

We decided to use only the most task-relevant item from the pilot evaluative measure here to

provide the simplest test of our hypotheses. An additional pre-test was carried out to check that this

‘creativity’ item could indeed be considered a measure of positive evaluation. Fifteen participants

were simply asked to rate the concept of creativity for valence on a scale anchored with negative at -


5, with neutral at 0 and with positive at +5. Creativity was significantly perceived to be positive

compared to the neutral midpoint: t (14) = 8.12, p < .001, (M= 3.53).

4

Although there appears to be outgroup favoritism in the multiple unrelated alternatives condition,

this is evidently more attributable to the lowering of ingroup evaluations than to an increase in

outgroup evaluations.

5

The analysis was also computed just on the creativity item and revealed exactly the same pattern, F

(1, 52) = 2.91, p = .063.


Table 1.

Evaluations as a function of categorization (Experiment 1).

Task

Control Multiple unrelated alternatives

Ingroup

7.27

(.961)

5.41

(2.15)

Outgroup

5.60

(1.72)

6.24

(1.64)

Intergroup Bias

1.67

(1.67)

-.823

(3.12)

Note: Higher numbers indicate more favourable evaluations for ingroup and outgroup evaluations,

but for intergroup bias higher numbers indicate more ingroup favoring bias. Standard deviations are

shown in parentheses.


Table 2.

Evaluations as a function of categorization (Experiment 2).

Task

Control

Multiple

Unrelated

alternatives

Multiple

Related alternatives

Ingroup

6.45

(1.11)

5.97

(1.73)

6.74

(.904)

Outgroup

6.16

(1.18)

6.62

(1.59)

6.35

(1.36)

Intergroup Bias

.249

(1.26)

-.650

(1.38)

.338

(1.18)

Note: Higher numbers indicate more favourable evaluations for ingroup and outgroup evaluations

but for intergroup bias higher numbers indicate more ingroup favoring bias. Standard deviations are

shown in parentheses.

considering multiple criteria for social categorization

Documents